arXiv subjects

Lucas C. Cordeiro

Publications and source records attributed to Lucas C. Cordeiro.

At least 19 recordsLinked to original sources

arXiv2026

Floating-Point Neural Network Verification at the Software Level

The behaviour of neural network components must be proven correct before deployment in safety-critical systems. Unfortunately, existing neural network verification techniques cannot certify the absence of faults at the software level. In this paper, we show how to specify and verify that neural networks are safe, by explicitly reasoning about their floating-point implementation. In doing so, we construct NeuroCodeBench 2.0, a benchmark comprising 912 neural network verification examples that cover activation functions, common layers, and full neural networks of up to 170K parameters. Our verification suite is written in plain C and is compatible with the format of the International Competition on Software Verification (SV-COMP). Thanks to it, we can conduct the first rigorous evaluation of eight state-of-the-art software verifiers on neural network code. The results show that existing automated verification tools can correctly solve an average of 11% of our benchmark, while producing around 3% incorrect verdicts. At the same time, a historical analysis reveals that the release of our benchmark has already had a significantly positive impact on the latter.

cs.SE↗

arXiv2026

Partial Contracts Suffice: Sound, LLM-Inferred Regression Verification

Software evolves continuously, yet ensuring that a patch preserves intended behavior without re-verifying an entire codebase remains difficult. Regression verification addresses this problem, but existing techniques require expensive whole-program reasoning or rely on manually written specifications that are rarely available in practice. We present the first contract-based regression verification tool. Contract soundness is ensured by proving all function versions match the behavior. The contract then verifies program flow via assume-guarantee. We ask whether a partial, caller-sufficient contract, rather than a full behavioral specification, is enough. On Frama-C-Problems we strengthen each inferred contract past what the caller needs and measure how much tighter it becomes. It barely moves: for most targets in every model the caller-sufficient contract is already the tightest the loop reaches, and our tightness comparator rates the partial and strengthened contracts equivalent for the large majority of targets it can compare. Partial-spec contracts thus capture nearly all the attainable tightness, so stopping at caller-sufficiency costs almost nothing. The regression check underneath is sound: on the third-party EqBench-C suite it never fabricates an equivalence, returning zero false proofs and reporting an unprovable difference instead. It also surfaced nine pairs that EqBench mislabels as equivalent, more than a concurrent tool reports. The contracts themselves are inferred automatically from the checker's own counterexamples, with no separate specification step; on Frama-C-Problems and the ANSSI X509 parser this reaches a verification rate comparable to tools AutoSpec and Preguss, while a passing result certifies at least as strong a property, which we call \emph{safety-preserving conditional equivalence}: enforcement plus caller-sufficiency.

cs.SE↗

arXiv2026

Neuro-Symbolic Software Verification: Hyper-charging Local Language Models with Symbolic Reasoning at Scale

Loop invariant synthesis remains a central and pivotal bottleneck in formal software verification. Recent LLM-based Neuro-Symbolic tools have achieved impressive solve rates. However, these tools rely on proprietary, often expensive cloud APIs, which constitute a hurdle for privacy-sensitive industrial deployments where the source code cannot leave the organisation or where cost is a factor. We present VerIbmc, a neuro-symbolic pipeline that pairs symbolic invariant generation with locally deployable open-weight language models with the ESBMC verification tool. Our pipeline combines a deterministic symbolic invariant synthesis phase with an iterative LLM refinement loop driven by structured verifier feedback. In addition, we provide two types of pipelines that differ in their prompting strategy: Chain-of-Thought vs. Tree-of-Thought. We conduct an extensive experimental evaluation with five open-weight models (ranging from 7B to 120B parameters) across five benchmark families comprising of 520 problems (499 after excluding 21 with unavoidable overflow). Overall, the best single configuration (GPT-OSS-120B) solves 431 of 499 problems (86.4%). Additionally, on the four benchmark suites shared with the strongest cloud-API tools, VerIbmc is competitive running only on a single local machine. The evaluation shows symbolic invariant synthesis solves 75 problems without any LLM call and yields up to +35 additional problems for the weakest model. Importantly, all inference runs entirely on a single local machine using open-weight models -- no cloud API or proprietary model is required. Overall, we demonstrate that a neuro-symbolic approach based on LLMs can be used effectively for invariant synthesis in a privacy-preserving and energy-efficient manner, without having to resort to expensive proprietary frontier models locked behind APIs.

cs.SE↗

arXiv2026

ConVer: Using Contracts and Loop Invariant Synthesis for Scalable Formal Software Verification

Formal verification of large C programs is impeded by state-space explosion: Bounded Model Checking (BMC) tools must encode the entire state space up to the predetermined bound by unrolling all nested constructs. We present ConVer, a top-down compositional verification tool. Given a C program with a top-level assertion, ConVer decomposes verification top-down: it uses a large language model (LLM) to synthesise function contracts from the system property, then alternates system-level and function-level checks in a CEGAR-CEGIS loop, refining contracts whenever a check fails via SMART ICE learning. We evaluate ConVer on four benchmark suites of increasing difficulty and against other state-of-the-art (SOTA) tools. On the Frama-C benchmark of 45 simple C programs, ConVer achieves 82-96% verification success across three LLM backends, with 93-95% of converged programs requiring only a single CEGAR-CEGIS iteration. On the X.509 parser benchmark (6~programs) and LF2C-Simple suite (17 programs), ConVer achieves 33-50% and 82-88% success respectively. On the VerifyThis suite of 11 recursive and loop-intensive programs, the Pre-Abstraction strategy achieves 55-64% success. In addition, we present ESBMC-LF a preprocessor tool that converts LF models to C while preserving the properties of the LF files, enabling ConVer to verify them. We transpile the LF Verifier Benchmarks using ESBMC-LF to C; we denote those LF-Hard. We show that ConVer successfully verifies 67% of LF-Hard benchmarks overall.

cs.SE↗

arXiv2026

FedSurrogate: Backdoor Defense in Federated Learning via Layer Criticality and Surrogate Replacement

Federated Learning remains highly susceptible to backdoor attacks--malicious clients inject targeted behaviours into the global model. Existing defenses suffer from substantial false-positive rates under realistic non-independent and identically distributed (non-IID) data, incorrectly flagging benign clients and degrading model accuracy even when adversaries are correctly identified. We present FedSurrogate, a novel backdoor defense that addresses this limitation by combining bidirectional gradient alignment filtering with layer-adaptive anomaly detection. FedSurrogate performs selective clustering on security-critical layers identified via directional divergence analysis, concentrating the detection signal on a low-dimensional subspace. A bidirectional soft-filtering stage screens trusted clients for residual contamination while rescuing false positives from suspects, substantially reducing misclassifications under heterogeneous conditions. Rather than removing confirmed malicious updates, FedSurrogate replaces them with downscaled surrogate updates from structurally similar benign clients, preserving gradient diversity while neutralising adversarial influence. Extensive evaluations demonstrate that FedSurrogate maintains false-positive rates below 10% across all datasets and attack types, compared to 31-32% for the nearest comparably effective baseline, while achieving superior main-task accuracy and maintaining attack success rates below 2.1% across all tested datasets and attack types under challenging non-IID settings.

cs.CR↗

arXiv2026

Specification Vibing for Automated Program Repair

Large language model (LLM)-driven automated program repair (APR) has advanced rapidly, but most methods remain code-centric: they directly rewrite source code and thereby risk hallucinated, behaviorally inconsistent fixes. This limitation suggests the need for an alternative repair paradigm that relies on a representation more accessible to LLMs than raw code, enabling more accurate understanding, analysis, and alignment during repair. To address this gap, we propose VibeRepair, a specification-centric APR technique that treats repair as behavior-specification repair rather than ad-hoc code editing. VibeRepair first translates buggy code into a structured behavior specification that captures the program's intended runtime behavior, then infers and repairs specification misalignments, and finally synthesizes code strictly guided by the corrected behavior specification. An on-demand reasoning component enriches hard cases with program analysis and historical bug-fix evidence while controlling cost. Across Defects4J and real-world benchmarks and multiple LLMs, VibeRepair demonstrates consistently strong repair effectiveness with a significantly smaller patch space. On Defects4J v1.2, VibeRepair correctly repairs 174 bugs, exceeding the strongest state-of-the-art baseline by 28 bugs, which corresponds to a 19% improvement. On Defects4J v2.0, it repairs 178 bugs, outperforming prior approaches by 33 bugs, representing a 23% improvement. Evaluations on real-world benchmarks collected after the training period of selected LLMs further confirm its effectiveness and generalizability. By centering repair on explicit behavioral intent, VibeRepair reframes APR for the era of "vibe" coding: make the behavior sing, and the code will follow.

cs.SE↗

arXiv2025

GPT, But Backwards: Exactly Inverting Language Model Outputs

The task of reconstructing unknown textual inputs to language models is a fundamental auditing primitive that allows us to assess the model's vulnerability to a range of security issues, including stealing hidden system prompts, detecting backdoors, and leaking private data. Existing inversion works assume access to differing levels of information (e.g. requiring input-output examples, the model parameters, intermediate activations or output logits) but oftentimes fail to fully reconstruct the desired input. In this paper, we present the Sparse One-hot Discrete Adam (SODA) algorithm, a search-based inversion method that can accurately reconstruct the input text, given white-box access to the language model and its output. Our experiments demonstrate for the first time that exact language model inversion is possible on both natural language and random inputs. Indeed, SODA achieves respectively 98% and 79% reconstruction rates on inputs with lengths up to 10 tokens. Furthermore, we show that input length and vocabulary size have a far greater impact on the probability of a successful reconstruction than the size of the language model itself, thus allowing us to scale to models from 33M to 3B parameters.

cs.LG↗

arXiv2025

Generative transformations and patterns in LLM-native approaches for software verification and falsification

The emergence of prompting as the dominant paradigm for leveraging Large Language Models (LLMs) has led to a proliferation of LLM-native software, where application behavior arises from complex, stochastic data transformations. However, the engineering of such systems remains largely exploratory and ad-hoc, hampered by the absence of conceptual frameworks, ex-ante methodologies, design guidelines, and specialized benchmarks. We argue that a foundational step towards a more disciplined engineering practice is a systematic understanding of the core functional units--generative transformations--and their compositional patterns within LLM-native applications. Focusing on the rich domain of software verification and falsification, we conduct a secondary study of over 100 research proposals to address this gap. We first present a fine-grained taxonomy of generative transformations, abstracting prompt-based interactions into conceptual signatures. This taxonomy serves as a scaffolding to identify recurrent transformation relationship patterns--analogous to software design patterns--that characterize solution approaches in the literature. Our analysis not only validates the utility of the taxonomy but also surfaces strategic gaps and cross-dimensional relationships, offering a structured foundation for future research in modular and compositional LLM application design, benchmarking, and the development of reliable LLM-native systems.

cs.SE↗

arXiv2025

UnitTenX: Generating Tests for Legacy Packages with AI Agents Powered by Formal Verification

This paper introduces UnitTenX, a state-of-the-art open-source AI multi-agent system designed to generate unit tests for legacy code, enhancing test coverage and critical value testing. UnitTenX leverages a combination of AI agents, formal methods, and Large Language Models (LLMs) to automate test generation, addressing the challenges posed by complex and legacy codebases. Despite the limitations of LLMs in bug detection, UnitTenX offers a robust framework for improving software reliability and maintainability. Our results demonstrate the effectiveness of this approach in generating high-quality tests and identifying potential issues. Additionally, our approach enhances the readability and documentation of legacy code.

cs.SE↗

arXiv2025

ReqInOne: A Large Language Model-Based Agent for Software Requirements Specification Generation

Software Requirements Specification (SRS) is one of the most important documents in software projects, but writing it manually is time-consuming and often leads to ambiguity. Existing automated methods rely heavily on manual analysis, while recent Large Language Model (LLM)-based approaches suffer from hallucinations and limited controllability. In this paper, we propose ReqInOne, an LLM-based agent that follows the common steps taken by human requirements engineers when writing an SRS to convert natural language into a structured SRS. ReqInOne adopts a modular architecture by decomposing SRS generation into three tasks: summary, requirement extraction, and requirement classification, each supported by tailored prompt templates to improve the quality and consistency of LLM outputs. We evaluate ReqInOne using GPT-4o, LLaMA 3, and DeepSeek-R1, and compare the generated SRSs against those produced by the holistic GPT-4-based method from prior work as well as by entry-level requirements engineers. Expert evaluations show that ReqInOne produces more accurate and well-structured SRS documents. The performance advantage of ReqInOne benefits from its modular design, and experimental results further demonstrate that its requirement classification component achieves comparable or even better results than the state-of-the-art requirement classification model.

cs.SE↗

arXiv2025

Supporting Software Formal Verification with Large Language Models: An Experimental Study

Formal methods have been employed for requirements verification for a long time. However, it is difficult to automatically derive properties from natural language requirements. SpecVerify addresses this challenge by integrating large language models (LLMs) with formal verification tools, providing a more flexible mechanism for expressing requirements. This framework combines Claude 3.5 Sonnet with the ESBMC verifier to form an automated workflow. Evaluated on nine cyber-physical systems from Lockheed Martin, SpecVerify achieves 46.5% verification accuracy, comparable to NASA's CoCoSim, but with lower false positives. Our framework formulates assertions that extend beyond the expressive power of LTL and identifies falsifiable cases that are missed by more traditional methods. Counterexample analysis reveals CoCoSim's limitations stemming from model connection errors and numerical approximation issues. While SpecVerify advances verification automation, our comparative study of Claude, ChatGPT, and Llama shows that high-quality requirements documentation and human monitoring remain critical, as models occasionally misinterpret specifications. Our results demonstrate that LLMs can significantly reduce the barriers to formal verification, while highlighting the continued importance of human-machine collaboration in achieving optimal results.

cs.SE↗

arXiv2025

ESBMC v7.6: Enhanced Model Checking of C++ Programs with Clang AST

This paper presents Efficient SMT-Based Context-Bounded Model Checker (ESBMC) v7.6, an extended version based on previous work on ESBMC v7.3 by K. Song et al. The v7.3 introduced a new Clang-based C++ front-end to address the challenges posed by modern C++ programs. Although the new front-end has demonstrated significant potential in previous studies, it remains in the developmental stage and lacks several essential features. ESBMC v7.6 further enhanced this foundation by adding and extending features based on the Clang AST, such as 1) exception handling, 2) extended memory management and memory safety verification, including dangling pointers, duplicate deallocation, memory leaks and rvalue references and 3) new operational models for STL updating the outdated C++ operational models. Our extensive experiments demonstrate that ESBMC v7.6 can handle a significantly broader range of C++ features introduced in recent versions of the C++ standard.

cs.LO↗

arXiv2025

CASTLE: Benchmarking Dataset for Static Code Analyzers and LLMs towards CWE Detection

Identifying vulnerabilities in source code is crucial, especially in critical software components. Existing methods such as static analysis, dynamic analysis, formal verification, and recently Large Language Models are widely used to detect security flaws. This paper introduces CASTLE (CWE Automated Security Testing and Low-Level Evaluation), a benchmarking framework for evaluating the vulnerability detection capabilities of different methods. We assess 13 static analysis tools, 10 LLMs, and 2 formal verification tools using a hand-crafted dataset of 250 micro-benchmark programs covering 25 common CWEs. We propose the CASTLE Score, a novel evaluation metric to ensure fair comparison. Our results reveal key differences: ESBMC (a formal verification tool) minimizes false positives but struggles with vulnerabilities beyond model checking, such as weak cryptography or SQL injection. Static analyzers suffer from high false positives, increasing manual validation efforts for developers. LLMs perform exceptionally well in the CASTLE dataset when identifying vulnerabilities in small code snippets. However, their accuracy declines, and hallucinations increase as the code size grows. These results suggest that LLMs could play a pivotal role in future security solutions, particularly within code completion frameworks, where they can provide real-time guidance to prevent vulnerabilities. The dataset is accessible at https://github.com/CASTLE-Benchmark.

cs.CR↗

arXiv2025

Vulnerability Detection: From Formal Verification to Large Language Models and Hybrid Approaches: A Comprehensive Overview

Software testing and verification are critical for ensuring the reliability and security of modern software systems. Traditionally, formal verification techniques, such as model checking and theorem proving, have provided rigorous frameworks for detecting bugs and vulnerabilities. However, these methods often face scalability challenges when applied to complex, real-world programs. Recently, the advent of Large Language Models (LLMs) has introduced a new paradigm for software analysis, leveraging their ability to understand insecure coding practices. Although LLMs demonstrate promising capabilities in tasks such as bug prediction and invariant generation, they lack the formal guarantees of classical methods. This paper presents a comprehensive study of state-of-the-art software testing and verification, focusing on three key approaches: classical formal methods, LLM-based analysis, and emerging hybrid techniques, which combine their strengths. We explore each approach's strengths, limitations, and practical applications, highlighting the potential of hybrid systems to address the weaknesses of standalone methods. We analyze whether integrating formal rigor with LLM-driven insights can enhance the effectiveness and scalability of software verification, exploring their viability as a pathway toward more robust and adaptive testing frameworks.

cs.SE↗

arXiv2025

SecureFalcon: Are We There Yet in Automated Software Vulnerability Detection with LLMs?

Software vulnerabilities can cause numerous problems, including crashes, data loss, and security breaches. These issues greatly compromise quality and can negatively impact the market adoption of software applications and systems. Traditional bug-fixing methods, such as static analysis, often produce false positives. While bounded model checking, a form of Formal Verification (FV), can provide more accurate outcomes compared to static analyzers, it demands substantial resources and significantly hinders developer productivity. Can Machine Learning (ML) achieve accuracy comparable to FV methods and be used in popular instant code completion frameworks in near real-time? In this paper, we introduce SecureFalcon, an innovative model architecture with only 121 million parameters derived from the Falcon-40B model and explicitly tailored for classifying software vulnerabilities. To achieve the best performance, we trained our model using two datasets, namely the FormAI dataset and the FalconVulnDB. The FalconVulnDB is a combination of recent public datasets, namely the SySeVR framework, Draper VDISC, Bigvul, Diversevul, SARD Juliet, and ReVeal datasets. These datasets contain the top 25 most dangerous software weaknesses, such as CWE-119, CWE-120, CWE-476, CWE-122, CWE-190, CWE-121, CWE-78, CWE-787, CWE-20, and CWE-762. SecureFalcon achieves 94% accuracy in binary classification and up to 92% in multiclassification, with instant CPU inference times. It outperforms existing models such as BERT, RoBERTa, CodeBERT, and traditional ML algorithms, promising to push the boundaries of software vulnerability detection and instant code completion frameworks.

cs.CR↗

arXiv2025

Neural Network Verification is a Programming Language Challenge

Neural network verification is a new and rapidly developing field of research. So far, the main priority has been establishing efficient verification algorithms and tools, while proper support from the programming language perspective has been considered secondary or unimportant. Yet, there is mounting evidence that insights from the programming language community may make a difference in the future development of this domain. In this paper, we formulate neural network verification challenges as programming language challenges and suggest possible future solutions.

cs.PL↗

arXiv2024

How secure is AI-generated Code: A Large-Scale Comparison of Large Language Models

This study compares state-of-the-art Large Language Models (LLMs) on their tendency to generate vulnerabilities when writing C programs using a neutral zero-shot prompt. Tihanyi et al. introduced the FormAI dataset at PROMISE'23, featuring 112,000 C programs generated by GPT-3.5-turbo, with over 51.24% identified as vulnerable. We extended that research with a large-scale study involving 9 state-of-the-art models such as OpenAI's GPT-4o-mini, Google's Gemini Pro 1.0, TII's 180 billion-parameter Falcon, Meta's 13 billion-parameter Code Llama, and several other compact models. Additionally, we introduce the FormAI-v2 dataset, which comprises 331 000 compilable C programs generated by these LLMs. Each program in the dataset is labeled based on the vulnerabilities detected in its source code through formal verification, using the Efficient SMT-based Context-Bounded Model Checker (ESBMC). This technique minimizes false positives by providing a counterexample for the specific vulnerability and reduces false negatives by thoroughly completing the verification process. Our study reveals that at least 62.07% of the generated programs are vulnerable. The differences between the models are minor, as they all show similar coding errors with slight variations. Our research highlights that while LLMs offer promising capabilities for code generation, deploying their output in a production environment requires proper risk assessment and validation.

cs.CR↗

arXiv2024

Dynamic Intelligence Assessment: Benchmarking LLMs on the Road to AGI with a Focus on Model Confidence

As machine intelligence evolves, the need to test and compare the problem-solving abilities of different AI models grows. However, current benchmarks are often simplistic, allowing models to perform uniformly well and making it difficult to distinguish their capabilities. Additionally, benchmarks typically rely on static question-answer pairs that the models might memorize or guess. To address these limitations, we introduce Dynamic Intelligence Assessment (DIA), a novel methodology for testing AI models using dynamic question templates and improved metrics across multiple disciplines such as mathematics, cryptography, cybersecurity, and computer science. The accompanying dataset, DIA-Bench, contains a diverse collection of challenge templates with mutable parameters presented in various formats, including text, PDFs, compiled binaries, visual puzzles, and CTF-style cybersecurity challenges. Our framework introduces four new metrics to assess a model's reliability and confidence across multiple attempts. These metrics revealed that even simple questions are frequently answered incorrectly when posed in varying forms, highlighting significant gaps in models' reliability. Notably, API models like GPT-4o often overestimated their mathematical capabilities, while ChatGPT-4o demonstrated better performance due to effective tool usage. In self-assessment, OpenAI's o1-mini proved to have the best judgement on what tasks it should attempt to solve. We evaluated 25 state-of-the-art LLMs using DIA-Bench, showing that current models struggle with complex tasks and often display unexpectedly low confidence, even with simpler questions. The DIA framework sets a new standard for assessing not only problem-solving but also a model's adaptive intelligence and ability to assess its limitations. The dataset is publicly available on the project's page: https://github.com/DIA-Bench.

cs.AI↗